1

Exploring Consumer Point-Of-Purchase Dynamics

Combining Virtual Reality and Choice Modeling in Consumer Purchase Analyses

Bernhard Treiber & Stephen P. Needel

Mulţumim, şi pe această cale,

autorilor acestui material, pentru

generoasa permisiune de a-l folosi

în scopuri didactice

2

Integration of two new approaches to consumer research

Choice-Based Conjoint Analysis

• involves the use of designed hypothetical choice situations

• measures consumer choices

• predicts their choices in new situations

Virtual Shopping Systems

• consist of a virtual „shelf“ or „store“ model and dy-namic virtual products

• can represent actual or hypothetical in-store decision tasks for con-sumers

• allows simulation of critical aspects of consumer shopping behavior

3

Discrete Choice Modeling

• Developed in parallel by economists and cognitive psychologists in late 1960s

• Statistical estimation techniques: McFadden, Louviere, Ben-Akiva et al.

• Wide applications:

– new product development

– positioning

– demand forecasts

– pricing research

– etc.

4

Discrete Choice Experiments

• Choice scenarios are constructed on the basis of fractional factorial designs, as a systematic variation of critical product attributes and their levels.

• Each respondent is shown - one at a time - the choice scenarios.

• Each scenario specifies the products from which the respondent is to choose.

• For each scenario, the respondent provides a behavioral intention response.

5

6

Discrete Choice Modeling

• Choice responses are decomposed into a set of part-worth utilities measuring the impact of each product attribute and its levels.

• Modeling is done through a particular type of statistical analysis (logit and probit analysis).

• Analysis is mostly done at the aggregate level (assuming that all consumers have the same preferences).

• Several software systems (e.g. SAS, CBC, LIMDEP) are available to run these analyses.

7

Discrete Choice Modeling Outcomes

• Quantitative decom-position of choice responses:

part-worth utilities measure the impact of each product attribute and its levels

• Computer-based Market Simulator:

will convert utility data into market „shares“ for different products under different competitive scenarios

8

Choice Tasks: Some Restrictions

• (mostly) limited number of options

• static depiction of choice alternatives

• predefined choice-consideration sets

• product information equalized across choice options

• de-contextualized choice setting

• „either-or“ choices

9

Choice Tasks: Desired Improvements

• (mostly) limited number of options

• static depictions of choice alternatives

• pre-defined choice-consideration sets

• „equal information“

• de-contextualized choice setting

• „either-or“ choices

• Many more option choices admitted

• more flexibility in displaying options

• respondent chooses rele-vant subsets of options

• „information inequalities“

• choice tasks in „natural“ settings

• more complex measures of choice behavior

Virtual Reality Applications

10

Virtual Store / Shopping Systems

• Developed in early 1990

• based on ‚virtual reality‘ technology

• several systems now available

• used for diverse applications:

– category management, shelf optimization, consumer research, store planning, package design, pricing

• major differences:

– focus: from entire stores to individual products

– technology: from VR headsets to standard PCs

– applications: strictly academic to fully commercialized

– project budgets

11

Virtual shopping systems Differences in Focus:

from individual products to entire stores

12

Virtual shopping systems Differences in Technology:

from standard PCs to VR headsets

13

VISIONARY SHOPPER

DEMONSTRATION

14

Visionary Shopper: a case example

• Main Features: – analytic unit: „products-in-shelves“

– display capacity: 300 products per shelf

– software: WINDOWS-98-based

– interaction medium: touch-screen

– project costs: low to moderate

– adoption: worldwide

• Origin: Harvard Business School

• Validation: positive

15

QUALITY

Baby Napkins: the German market

48 Items / DM 19,99

44 Items / DM 23,99

60 Items / DM 32,99

PRICE

16

QUALITY

Strategic Options for the Middle Position

48 Items / DM 19,99

44 Items / DM 23,99

60 Items / DM 32,99

PRICE

17

PACKAGE CHANGE

Retail Prices

• 23,99 / 56

• 24,99 / 52

• 25,99 / 56

• 28,99 / 62

Price Per Item

• 42,8 = „current“

• 48,1

• 46,3

• 46,7

PRICE CHANGE

“Old” “New”

18

PACKAGE CHANGE

Retail Price

• 32,99 / 70

• 32,99 / 62

Price Per Item

• 47,1 = „current“

• 53,2

PRICE CHANGE

“Old” “New”

19

QUALITY

The German market for baby napkins:

Repositioning Options for „Player A“ and „Player B“

42,8 46,3 46,7 48,1 PRICE / ITEM

47,1 53,2

Fixies:

Pampers:

„current“

„current“

20

Experimental Design

PACKAGE

• „old“

• „new“

PRICE / ITEM

• 42,8

• 46,3

• 46,7

• 48,1

• 47,1

• 53,2

FIXIES

PA1

PA2

PR1

PR2

PR3

PR4

PAMPERS

PA1

PA2

PR5

PR6

OTHERS

un-

changed

NONE

21

Design Considerations

2-Factor-Designs ( Package x Price):

• FIXIES: 2 x 4 = 8 test combinations

• PAMPERS: 2 x 2 = 4 test combinations

• FIXIES x PAMPERS: 8 x 4 = 32 test cells

Fractional Factorial Design:

only a subset of all possible test combinations selected and implemented: = 16 test cells

22

DESIGN CONSIDERATIONS

Experimental Design

• specifies critical combinations of Experimental Levels / Attributes („Choice Tasks“)

Visionary Shopper System

• puts choice tasks into the context of typical purchase environment („Baby napkin shelf in German supermarkets“)

• generates „new“ packaging through digital editing

• programs a total of 16 test shelves

23

24

TEST PROCEDURE

• 16 test cells split into 4 subsets of 4 cells each

• each respondent goes through one of these subsets (= 4 test cells)

• test category („Baby Napkins“) is shown together with two other categories („Wipes“, „Bandages“)

• each test cells is seen by n=90 respondents

• total sample: N=374 mothers of babies aged 6-16 months

25

TEST RUNS

Each respondent is taken through 3 product

categories on 4 shopping trips

• Trip 1: Bandages Napkins (#1 of 16) Wipes

• Trip 2: Bandages Napkins (#2 of 16) Wipes

• Trip 3: Bandages Napkins (#3 of 16) Wipes

• Trip 4: Bandages Napkins (#4 of 16) Wipes

26

DATA COLLECTION:Fully automized, yielding a second-to-second

protocol of shopper behavior-in-category

• Time spent in category

• Time spent on individual products

• Close contact with individual products

• Sequence of product contacts

• Product contacts without purchase

• Purchases

– „Yes / No“

– Number of packs

• plus Follow-up interview

27

CHOICE MODELING

Purchase-related measures available

• for each of 32 products

• under 16 choice tasks

• representing all 32 possible test combinations

Input for Choice Modeling:

• Multinominal logit models are fitted to the data

• every attribute level is assigned a utility

• indicating relative importance of each level

28

CLIENT BENEFITSImproved Decision-Making Abilities

• Detailed quantitative understanding of how to best reposition own product line in view of likely competitor moves

• Encouraged to reposition own product line to a Hi-Price/Hi-Quality position (close to PAMPERS)

• When executed, sales, revenues, and profits were much improved,

• to the disadvantage of PAMPERS.

29

RESEARCH BENEFITS 1

• Complex consumer marketing issues translate more easily into equally complex research de-signs, avoiding test-practical oversimplifica-tions, short-cuts, and reductions in data collection.

• Choice modeling of shopper behavior and pur-chase decisions is now made possible at the individual level

– with much more granularity than ever before,

– for multiple indices of shopping behavior

30

RESEARCH BENEFITS 2

• Hypothetical „what-if“scenarios for retail, cate-gory, and product management can now be simulated for even large numbers of options, including

– likely competitor moves

– new product concepts

– newly created shopping environments.

• Traditional approaches for in-store research (e.g. Shelf Tests, Concealed Video-cameras) prove less efficient, consistent, and comparable across different test conditions.

31

ENDThank You.

32

Exploring Consumer Point-Of-Purchase Dynamics

Combining Virtual Reality and Choice Modeling in Consumer Purchase Analyses

• Bernhard Treiber & Stephen P. Needel

33

CATEGORIES

GROCERY HOUSEHOLD OTC PERSONAL CARE

Baby Food Air Fresheners Acid Relief Baby Wipes

Baby Formula Automobile Tires Allergy Relief Bar Soap

Baked Beans Batteries Analgesics Body Wash

Beer Bleach Athelete's Foot Contraceptives

Butter/Margarine Cleaners Bandages Deodorants

Candy Dinner Napkins Bladder Infection Diapers

Canned Pasta Fabric Softener Corn/Callous Disposable Razors

Canned Tuna Laundry Detergent Cough/Cold Incontinence

Carbonated Beverages Mops/Brooms Laxatives Insoles

Chewing Gum Paper Towels Menopause Relief Personal Lubricants

Cigarettes Plastic Wraps Nasal Sprays Sanitary Protection

Cookies Soap Pads Stomach Remedies Shampoos

Crackers Storage Bags Thermometers Sun Care

Frozen Dinners/Entrees Tissues Throat Lozenges Toothpaste

Hot Cereal Toilet Paper Vitamins/Minerals

Ice Cream Novelties Trash Bags Yeast Infection Remedies

Juice - Aseptic

Juice - Chilled

Juice - Frozen

Juice - Shelf Stable

Ketchup

Lunch Meat

Mayonnaise

Nuts

Packaged Dinners

Ready to Eat Cereal

Salad Dressing

Salty Snacks

Spirits

Tea

Yogurt

34

VALIDATION

0.75 0.8 0.85 0.9 0.95 1

CORRELATION

PAPER TOWELS

CARB BEV

LAXATIVES

RTE CEREAL

SHAMPOO

NASAL SPRAY

LAUNDRY DTG

SNACK NUTS

SALTY SNACKS

0.80

0.83

0.86

0.89

0.90

0.92

0.94

0.94

0.95

MARKET SHARE VS. SIMULATION SHARE

• When you ask people to shop normally, they do!

35

-6

-5

-4

-3

-2

-1

0

1

2

3

SIM

UL

AT

IO

N S

HA

RE

C

HA

NG

E

-6 -4 -2 0 2 4 KROGER SHARE CHANGE

121 SKU VS. 85 SKU SHARE CHANGESREVISED MARKET SHARES

VALIDATION

Visionary Shopper can also be real-world predictive